ses: Fix racy cleanup of /sys in remove_dev()

jcalvinowens · martinkpetersen · commit e120dcb6b2a1 · 2016-08-12T17:40:51.000-04:00
Currently we free the resources backing the enclosure device before we
call device_unregister(). This is racy: during rmmod of low-level SCSI
drivers that hook into enclosure, we end up with a small window of time
during which writing to /sys can OOPS. Example trace with mpt3sas:

  general protection fault: 0000 [#1] SMP KASAN
  Modules linked in: mpt3sas(-) &lt;...&gt;
  RIP: [&lt;ffffffffa0388a98&gt;] ses_get_page2_descriptor.isra.6+0x38/0x220 [ses]
  Call Trace:
   [&lt;ffffffffa0389d14&gt;] ses_set_fault+0xf4/0x400 [ses]
   [&lt;ffffffffa0361069&gt;] set_component_fault+0xa9/0xf0 [enclosure]
   [&lt;ffffffff8205bffc&gt;] dev_attr_store+0x3c/0x70
   [&lt;ffffffff81677df5&gt;] sysfs_kf_write+0x115/0x180
   [&lt;ffffffff81675725&gt;] kernfs_fop_write+0x275/0x3a0
   [&lt;ffffffff8151f810&gt;] __vfs_write+0xe0/0x3e0
   [&lt;ffffffff8152281f&gt;] vfs_write+0x13f/0x4a0
   [&lt;ffffffff81526731&gt;] SyS_write+0x111/0x230
   [&lt;ffffffff828b401b&gt;] entry_SYSCALL_64_fastpath+0x13/0x94

Fortunately the solution is extremely simple: call device_unregister()
before we free the resources, and the race no longer exists. The driver
core holds a reference over -&gt;remove_dev(), so AFAICT this is safe.

Signed-off-by: Calvin Owens &lt;calvinowens@fb.com&gt;
Reviewed-by: James Bottomley &lt;jejb@linux.vnet.ibm.com&gt;
Signed-off-by: Martin K. Petersen &lt;martin.petersen@oracle.com&gt;
diff --git a/drivers/scsi/ses.c b/drivers/scsi/ses.c
@@ -778,6 +778,8 @@ static void ses_intf_remove_enclosure(struct scsi_device *sdev)
 	if (!edev)
 		return;
 
+	enclosure_unregister(edev);
+
 	ses_dev = edev->scratch;
 	edev->scratch = NULL;
 
@@ -789,7 +791,6 @@ static void ses_intf_remove_enclosure(struct scsi_device *sdev)
 	kfree(edev->component[0].scratch);
 
 	put_device(&edev->edev);
-	enclosure_unregister(edev);
 }
 
 static void ses_intf_remove(struct device *cdev,
